okular search not working

[Dave Gilbert]

Load a PDF into okular, hit ctrl-F, type your search term - nothing happens except for the spinny icon keeping spinning.
I do notice that on the terminal I have had:

QMetaObject::invokeMethod: No such method Okular::Document::doContinueNextMatchSearch(void *,void *,int,int,QString,int,bool,QColor,bool)

This is on okular package 4:4.8.90-0ubuntu1 (0.14.90 okular) and also on project-neon's 0.15.70 okular (2+git20120721+r4029-1).

I'll link to the upstream bug report; they're asking for a test on rc2, because it's working on rc2 on some other platform.

ProblemType: Bug
DistroRelease: Ubuntu 12.10
Package: okular 4:4.8.90-0ubuntu1
ProcVersionSignature: Ubuntu 3.5.0-5.5-generic 3.5.0-rc7
Uname: Linux 3.5.0-5-generic x86_64
ApportVersion: 2.4-0ubuntu4
Architecture: amd64
Date: Sun Jul 22 17:33:06 2012
InstallationMedia: Kubuntu 12.10 "Quantal Quetzal" - Alpha amd64 (20120717)
ProcEnviron:
 LANGUAGE=en_GB:en
 TERM=xterm
 PATH=(custom, no user)
 LANG=en_GB.UTF-8
 SHELL=/bin/bash
SourcePackage: okular
UpgradeStatus: No upgrade log present (probably fresh install)

[Munzir Taha (منذر طه) (munzirtaha)]

Searching for a text does not find any result.

Reproducible: Always

Steps to Reproduce:
1. Open Okular
2. Edit -> Find...
3. Type a word
Actual Results:
Nothing found

Expected Results:
Any word in the document should be found

munzir@GT683R:~$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu quantal (development branch)
Release: 12.10
Codename: quantal
munzir@GT683R:~$ apt-cache policy okular
okular:
  Installed: 4:4.8.90-0ubuntu1
  Candidate: 4:4.8.90-0ubuntu1
  Version table:
 *** 4:4.8.90-0ubuntu1 0
        500 http://archive.ubuntu.com/ubuntu/ quantal/universe amd64 Packages
        100 /var/lib/dpkg/status

[Gilbertd+kde (gilbertd+kde)]

Same here, also 4:4.8.90-0ubuntu1 ubuntu quantal; Okular identifies itself as 0.14.90 , using KDE Development Platform 4.8.90 (4.8.90)

Just keeps spinning the rotating progress icon, never finishes, even on a couple of page doc.

[Albert Astals Cid (aacid)]

Please, attach a document with such a problem.

[Gilbertd+kde (gilbertd+kde)]

Created attachment 72647
www.kde.org/.../QtContributionLicenseAgreement-1-1.pdf

Try searching for anything in this (It's just the Contribution agreement of kde.org for qt); e.g.

ctrl-f
Agreement

never stops, shows feedback or anything.

[Albert Astals Cid (aacid)]

Works like a charm here, any change you can try with rc1 or rc2?

[Gilbertd+kde (gilbertd+kde)]

Still seems to happen for me with okular 0.15.70 on kde 4.8.97 (that I think is rc2) using the project-neon package build 2+git20120721+r4029-1~quantal1; this run is in a 32bit KVM guest, as opposed to my original run on 64bit host; so it's not word size or graphics card dependent.

[Albert Astals Cid (aacid)]

okular 0.15.70 is not rc2, if your package for rc2 contains that, tell your packager, rc2 is okular 0.14.97

[Gilbertd+kde (gilbertd+kde)]

OK, thanks - I don't have access to a specific rc2 set of .deb's - although if you have pointers I'd be interested.

Having said that, given that I'm able to recreate this on 0.14.90 prior to the rc's, and you're saying the 0.15.70 is after the rc's it would be surprising if it had been fixed just for the rc version.

(I probably won't have a chance to do another test until next weekend).

[Albert Astals Cid (aacid)]

Well, it works for me as i said in rc2

I'm not an ubuntu packager, the only thing i can say is you can try building from source tarball of the rc2 packages if you know how to do that.

[Albert Astals Cid (aacid)]

Adding the kubuntu packagers so they can see the fact that they decided to pckage rc2 is hurting our users

[Gilbertd+kde (gilbertd+kde)]

Albert:
  I've added an Ubuntu bug ( https://bugs.launchpad.net/ubuntu/+source/okular/+bug/1027657 )
that links back to this one so it can be tracked.

I think the only reason rc2 isn't packaged was a lack of resource.

Of course it's not obvious that it would work in rc2 either and it might be something more subtle.

[Gilbertd+kde (gilbertd+kde)]

One thing I hadn't previosly noticed (and don't know if is relevant); if I start okular on the command line and do the search I see:

QMetaObject::invokeMethod: No such method Okular::Document::doContinueNextMatchSearch(void *,void *,int,int,QString,int,bool,QColor,bool)

(both on the older and newer version I tried).

[Albert Astals Cid (aacid)]

Yes, this is very important.

Anyway i've confirmed with ubuntu developers that precise+okular from project neon can search and that quantal+okular from project neon can not search so it is a downstream (aka Ubuntu issue) and they'll have to deal with it since there's nothing i can fix in Okular itself

[Gilbertd+kde (gilbertd+kde)]

Ok thanks Albert.

[Albert Astals Cid (aacid)]

Thanks for caring about Okular, let's hope the kubuntu devels can find a fix soon.

[Albert Astals Cid (aacid)]

*** Bug 303933 has been marked as a duplicate of this bug. ***

[Albert Astals Cid (aacid)]

Rex, adding you since the duplicated 303933 affects fedora too, maybe you guys share some bad patch in Qt or gcc? People have confirmed it works fine in Arch gcc 4.7 and Qt 4.8.2

[Rdieter-math (rdieter-math)]

I can't reproduce on my own f17/x86_64 box (ie, it searches pdf's just fine). Is there a chance of a missing builld or runtime dependency or is okular just using facilities directly from poppler here?

[Albert Astals Cid (aacid)]

No poppler, it's just okular code.

The
QMetaObject::invokeMethod: No such method Okular::Document::doContinueNextMatchSearch(void *,void *,int,int,QString,int,bool,QColor,bool)
means something was very wrong on compilation/execution since that is never invoked, what is invoked is Okular::Document::doContinueNextMatchSearch(void *,void *,int,int,QString,int,bool,QColor,bool,int) in lines 1224 and 2956 of document.cpp so there is something wrong somewhere, i'd blame either gcc or qt, but to be honest i have no clue

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Rik Mills (rikmills)]

Seeing the same here in Quantal with Okular v0.14.90 and project neon with v0.15.70

[Rik Mills (rikmills)]

Forceably downgrading the qt4-x11 packages in Quantal from 4:4.8.2-0ubuntu4 to the Precise versions 4:4.8.1-0ubuntu4.2 seems to restore search functionality in Okular.

Breaks other things/depends, but seems the source of this issue?

[Rik Mills (rikmills)]

Bug still present when running new okular 4.9 release Quantal build.

[Lisandro Damián Nicanor Pérez Meyer (perezmeyer)]

This bug has hitted qt4-x11/4:4.8.2-2 but no qt4-x11/4:4.8.2-1. Apart from the patches between them (two very small and contained), the only difference seems to be GCC.

I'm currently rebuilding qt4-x11/4:4.8.2-1 with current GCC to see if it gets triggered.

[Rdieter-math (rdieter-math)]

Tracking downstream @
https://bugzilla.redhat.com/show_bug.cgi?id=846847

fedora users, please followup there.

[Lisandro Damián Nicanor Pérez Meyer (perezmeyer)]

(In reply to comment #19)
> This bug has hitted qt4-x11/4:4.8.2-2 but no qt4-x11/4:4.8.2-1. Apart from
> the patches between them (two very small and contained), the only difference
> seems to be GCC.

I forgot to mention that this is Debian ;)

The Debian bug is in <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=684177>

[Kevin-kofler (kevin-kofler)]

GCC is also the main difference between Fedora 17 and 18 (gcc-4.7.0-5.fc17 vs. gcc-4.7.1-5.fc18). So it may well be a regression in GCC.

[Lisandro Damián Nicanor Pérez Meyer (perezmeyer)]

Indeed, qt4-x11/4:4.8.2-1 was compiled with gcc-4.7.0-1 and qt4-x11/4:4.8.2-2 with gcc-4.7.1-2 (both Debian revisions).

[Lisandro Damián Nicanor Pérez Meyer (perezmeyer)]

The problem is clearly a gcc bug. It has gone with version 4.7.1-6 currently in Debian Sid, you may want to look for patches there, although the correct one could be between versions -2 and -6 inclusive.

Kinds regards, Lisandro.

[Albert Astals Cid (aacid)]

Good work :-)

Is any of you guys reporting this upstream to gcc?

[Lisandro Damián Nicanor Pérez Meyer (perezmeyer)]

(In reply to comment #25)
> Is any of you guys reporting this upstream to gcc?

I haven't spotted the right patch in gcc (nor I tried, to be honest). In other words, I don't have a clue what could have possibly changed.

[Albert Astals Cid (aacid)]

Well, but we have confirmed it's a gcc regression, no?

We can tell them and maybe they can find faster?

[Kevin-kofler (kevin-kofler)]

Comment #24 implies this is already fixed in current GCC, at least in Debian, but it probably got fixed by something from the GCC 4.7 branch. So I'm not sure reporting it to upstream GCC is useful.

[Albert Astals Cid (aacid)]

Oh, i can't read, i thought he had gone back in versions, not forward, ignore my suggestion then

[Rik Mills (rikmills)]

Yes, problem is fixed in the latest project neon using qt built with gcc 4.7.1-6ubuntu1

[Dave Gilbert (ubuntu-treblig)]

Added ubuntu gcc package in; since the consensus seems to be an unknown (possibly fixed)
gcc bug. Marking High since if it's producing broken packages that would be a bad thing.

[Matthias Klose (doko)]

fixed in gcc-4.7 4.7.1-6 and later

[Dave Gilbert (ubuntu-treblig)]

Well it's working on 4.9.0-0ubuntu2 on current Quantal.